FIOS + Vuze Friend Boost?

nweibley

Member
Joined
Apr 27, 2005
Messages
60
Location
Tampa/Gainesville, Fl
Just throwing this out there as a potential idea:

I don't know how many FIOS users we have on FG, and I don't know how many of them are Vuze users, but I'm curious if there'd be any interest in this.

Basically, apply the Vuze "friend boost" feature to a core of well connected FIOS users to distribute certian FG.com content. Since we have atypically fast and stable upstream throughput, if we were to peer together and seed to one-another with let's say 1/3rd your rated upstream bandwidth, I believe we could have a fairly significant positive impact on the availability and swarm speed of torrents from release. By preferentially peering to one-another and staying inside the core verizon network to form an early base of seeds I would suspect we would at least have some positive impact.

With 20 20Mbit/5Mbit FIOS users allocating a 1.67Mbit a piece to friend boosting one another, that would allow the network of "friends" to pull really quick d/ls of the source which they could seed at 3.3+ Mbit to the rest of the swarm. Consider that a pool of 20 5Mbit FIOS upstreams would contribute 100Mbit and that's not a shabby bit of bandwidth to help the process along. Of course with more FIOS users or 20Mbit symmetric lines it would be even better, but you get the gist.

Thoughts? Interest? Concerns? I'd love to give it a try if there are other FIOS users out there willing to take part.
 
Well, afaik Viper and others are on it with 3 100Mbit lines, so your idea would help a bit, but not as much as you might think....
 
I don't doubt that Viper has huge seeders based on the availability of the torrent right after it's been posted to the front page and only a few minutes old. I've been a part of that phase before and the bandwidth is impressive.

That said it could never hurt for your best 'residential' seeders to get a copy quickly, because they only help to further propagate and decentralize the data. Right now FIOS users have fast pipes (at least for consumer broadband in North America) but must compete with the massive number of regular users hitting the base seeds at the same time. A saturated 20/5 FIOS user could have a full 700MB data block in just over 4.5 minutes... it wouldn't take long for a core FIOS population to spread the data to one another in favored channels. IDK, I guess people still generally get stuff fairly quickly after the release (I'm usually done in around 20mins or less) but I'm just sorta musing if there's room for improvement and if a decentralized base of users can help, why not?

I guess also there'd be some burden of proof of FIOS ownership so it wouldn't just be people hopping in to get favored connections. speedtest.net results or something may suffice though, I suppose. I get 18.5/4.4 at 8pm to Texas (abt 800mi)


I guess it would all depend on how many FIOS users were on the forum and willing to seed for a while.
 
not a bad idea.im not around your parts but tend to seed top gear stuff from time to time....i had season 3 on seed mode for a while.....was surprised at the number of folks who were snatching it...:)



edit...hmm.....i tought i had some posts in here....well...seems ff remembering me user name and pass was a plus :D...ahem...konichiwa!
 
Assuming that this idea takes off, would AT&T's U-Verse work? It's quite similar to FiOS, but it's not fiber directly to the home...
 
In theory, anything that gets more chunks out to people with good upload quickly is good for the swarm, but in reality, I suspect that any improvement would be minimal. With so many people grabbing a new torrent at once, and a decent algorithm choosing what chunks to send, it takes very little time for everyone's upload to be saturated. The people with lots of upload should feel free to disabuse me of this theory if experience proves me wrong however. I haven't done any experiments to test it.

As for the usefulness of keeping things within the Verizon (or ATT with U-Verse) networks, it is really only helpful to the backbone providers, and then only when done on a massive, global scale. Since the people would become primary seeders even more quickly, and thereafter push out lots of data to everyone on the other networks, I think that even that advantage would be quickly overshadowed.

Basically, keeping things on one provider's network helps keep the backbones and cross-provider interconnects unclogged, but the reality is that the Internet backbone has plenty of bandwidth. It's only the last mile that is congested, and having a bunch of nodes at the edges of a couple networks does nothing to alleviate that.
 
Assuming that this idea takes off, would AT&T's U-Verse work? It's quite similar to FiOS, but it's not fiber directly to the home...

Yes, the principle stands for anyone with a particularly fat connection. Though it doesn't hurt to stay w/in your provider's network, but I digress.

The presumed benefit is that users with fat pipes can "outsmart" the normal BT peering algorithm via the friend boost option ensuring that the fastest users all have a copy to seed ASAP by using their own core network to distribute to one another (thus cutting down on the amount the main initial seeds will have to transmit to the high-bandwidth leechers anyways). If the proportions of bandwidth are allocated correctly "fat piped" users would in theory be able to simultaneously saturate their up and downstream channels. This would shorten the amount of time necessary for them to stop using the seed server's bandwidth and free them up to act as seeds for the regular broadband users. Instead the regular swarming algorithm keeps most FIOS users from connecting as peers and swapping with one another from the get go.


We'd need a bunch of fat-piped FIOS users to make an "oh hey, that worked" noticeable difference though, I'd imagine. And really I could see it going two ways:
FIOS users all get a copy of the data very quickly and help to decentralize the load fairly moderately or
FIOS uses all get a copy of the data quickly but their impact on the availability and swarm speed is negligible.

I'll have to think about it some more, but I do think there is potential for benefit if done correctly.
 
In theory, anything that gets more chunks out to people with good upload quickly is good for the swarm, but in reality, I suspect that any improvement would be minimal. With so many people grabbing a new torrent at once, and a decent algorithm choosing what chunks to send, it takes very little time for everyone's upload to be saturated. The people with lots of upload should feel free to disabuse me of this theory if experience proves me wrong however. I haven't done any experiments to test it.

As for the usefulness of keeping things within the Verizon (or ATT with U-Verse) networks, it is really only helpful to the backbone providers, and then only when done on a massive, global scale. Since the people would become primary seeders even more quickly, and thereafter push out lots of data to everyone on the other networks, I think that even that advantage would be quickly overshadowed.

Basically, keeping things on one provider's network helps keep the backbones and cross-provider interconnects unclogged, but the reality is that the Internet backbone has plenty of bandwidth. It's only the last mile that is congested, and having a bunch of nodes at the edges of a couple networks does nothing to alleviate that.

Yes, I agree with all of what you have said (hard not to since it is correct). What experience has shown me is that the first 5-10min of seeding is moderately effective but not full-tilt. Also, it's complicated to account for the ramp-up in demand on the torrent. If the FIOS users were able to have a full copy and propagate to new leechers before a massive influx it is possible that there would be benefit. I see what you are saying re: "If you already saturate your upstream without a full copy, what good does a full copy do you any quicker" however that's not accounting for the fact that FIOS users with even fatter downstream pipes would be done sucking bandwidth from the swarm earlier.

I'm not sure; I'm kind-of undecided if it would help... I don't think it would hurt but I do think it would require at least 100Mbit+ of total upstream before there was any perceptible effect.
 
FIOS by default is fast, and from what I've seen on pricing most have good speeds, so it would definitely help FIOS users to work together so at least THEY would be able to share the episode among themselves as fast as possible. But it's all well and good getting the episode fast, don't forget to keep seeding :)

We manage at least 4x100mbit on Sunday night, sometimes another 2x100mbit, all depends on who's feeling generous ;)
 
I seed over 12-20 GB an episode for the past few :) Fios is cool although I wish our routers were not as restrictive because we could probably get a bit more speed out of the connections.
 
Even still with the default Actiontec routers I'm usually able to fully saturate my upstream bandwidth seeding high-demand torrents.

It looks like we've got maybe 3 FTTH users in the thread so far. I'm not sure how many FIOS users are actually on the forum though, much less how many would be willing to use Vuze. Then again this thread doesn't seem much traffic so... :confused:

I guess I'll sit on it and see if more FIOS users show up. If not, oh well.
 
You don't need 100% of the file to be helpful and seed. You can help out by even just seeding a few pieces of the file fast.

Most of our boxes run uTorrent anyway.
 
FIOS user here, but I won't use anything else besides utorrent....

And not to mention I'm like a season behind in my TG and FG :(
 
Top